canonical
Need longer timeout and retries when probing instance for mount support
When an instance is finished booting, it's possible it's very busy starting and running services. As such, when multipassd probes the instance to determine if mount support is enabled, etc., the instance may be so busy that the probing may time out. We should increase the timeout and/or add some retries.
Original message: I am still having this issue with
multipass 1.12.2
multipassd 1.12.2
Originally posted by @sed-i in https://github.com/canonical/multipass/issues/2986#issuecomment-1724664169
Hi @sed-i!
After looking through the log, I see that there was an ssh timeout when the mount handling code was trying to probe the instance to make sure related software was installed. Since it got a failure, it removed the mount. Does trying to add the mount again work?
Does trying to add the mount again work?
Absolutely. More often than not, after the host reboots I need to manually mount. And it works right away.
Ok. I only see one time in the log of where it was removed. Would it possible to get more cases of this so I can be sure of the exact scenario when this occurs? Also, do you happen to have many services running inside the instances that bog it down and cause it to not respond in a timely manner? I'm just wondering if these instances are busy on start.
do you happen to have many services running inside the instances that bog it down and cause it to not respond in a timely manner?
Yes, I think so: every time the vm boots up, juju brings back all the apps, so I imagine it could be resource intensive.
Ok, so we should probably add a longer timeout and/or be a bit more persistent in trying again a few times.
